a) it is just obfuscating things.  All of the issue34 work happens at
the stage of iteration when tags have already long been processed.

Understood, I see then that \tag etc. aren't really related to #34.

If I think of any other things, I will try again (and this time I will write in a way regtest-inclusion friendly). But I can see a lot has already been covered.

Thanks,
Yoshi



On 2026-10-20 14:43, David Kastrup wrote:
Yoshiaki Onishi <[email protected]> writes:

Looking at the regtests to which you have provided the link, I was
wondering if the following code using \tag might be worth testing
out. It uses \tag, \keepWithTag, and \removeWithTag:

%% B E G I N %%

% Grace note example made by Yoshiaki Onishi (October 2026) - for David Kastrup
\version "2.26.0"
withGrace = {
 \tag #'grace { c'1 \time 3/4 \grace {c'16 } e'2.}
 \tag #'withoutGrace { c'1 \time 3/4 e'2.}
 \tag #'howItShouldLook { c'1 \time 3/4 \grace {s16} e'2.}
}
\score {
 \new StaffGroup
 <<
  \keepWithTag #'grace \withGrace
%%%%% Uncomment one of the following three lines %%%%%
  %\keepWithTag #'withoutGrace \withGrace
  %\removeWithTag #'(grace howItShouldLook) \withGrace
  %\keepWithTag #'howItShouldLook \withGrace
 >>
}
%% E N D %%

I don't think that makes sense since

a) it is just obfuscating things.  All of the issue34 work happens at
the stage of iteration when tags have already long been processed.

b) "% Uncomment one of the following three lines %" is nothing we can do
in a regtest.  You would need to create three separate regtests for
that, one for each line, and then there is no need to use tags but you
can just write out the code that remains after tag processing.

Alternatively you'd need to write three different scores, one for each
of the uncommented lines.  Point a) would still apply: you could just
write out the versions tag processing would produce and get a clearer
view on what is being tested.

In addition, all three versions actually look the same even in 2.27.4.
Presumably this happens to already be fixed (at least visually) due to

commit 8c482233f024f658b159d630607f8a57a36dfdf6
Author: David Kastrup <[email protected]>
Date:   Sun Sep 20 01:27:51 2026 +0200

    Only emit one time signature per main time step

Time signature changes after grace notes are undesired. In case the desired time signature changed after the fact (can happen in polymetric situations),
    the existing grob is rewritten in order to retain alignment.

--
Yoshiaki Onishi
https://github.com/yoshiakionishi/lilypond-snippets

Reply via email to